Wireless network intelligent planning method and device, equipment, storage medium and product

Abstract

The invention discloses a wireless network intelligent planning method and device, equipment, a storage medium and a product, and the method comprises the steps: obtaining a three-dimensional grid model and electromagnetic parameter distribution of a to-be-planned environment, building the three-dimensional grid model based on three-dimensional point cloud data with material information, and employing a ray tracing technology to carry out the electromagnetic parameter distribution of the to-be-planned environment; and performing wireless signal propagation simulation on the three-dimensional grid model to obtain a signal propagation data set, inputting the signal propagation data set into a wireless network intelligent planning model, and predicting the position of the wireless access point through the wireless network intelligent planning model to obtain the position of the wireless access point. According to the embodiment of the invention, the deployment efficiency of the position of the wireless access point is improved.

Description

Technical Field

[0001] The present application relates to the field of wireless network technology, and in particular to a method, apparatus, device, storage medium, and product for intelligent planning of a wireless network. Background Art

[0002] With the advent of the 5G era, users are increasingly demanding higher performance from wireless networks, moving beyond wide-area outdoor coverage to include high-speed, low-latency, and highly reliable connections in indoor environments. For example, the explosive growth of smart terminals, IoT devices, and various mobile applications is placing higher demands on indoor wireless networks in places like office buildings, shopping malls, hospitals, and airports.

[0003] Existing technologies primarily assess indoor signal coverage using free-space path loss models, wall attenuation models, or empirical corrections. In other words, engineers conduct field surveys and draw indoor layouts. Based on these layouts, they predict indoor signal attenuation using free-space path loss and wall attenuation models. Then, based on their experience, they determine the initial placement of wireless access points (APs). RF simulation tools are then used to predict signal coverage, signal strength, and interference at the AP locations. Finally, the AP locations are adjusted based on the predicted results.

[0004] However, when the environment is complex, engineers need to manually adjust the location of wireless access points multiple times to find a more suitable access point, which reduces the deployment efficiency of the wireless access point location. Summary of the Invention

[0005] The embodiments of the present application provide a wireless network intelligent planning method, apparatus, device, storage medium and product, which can improve the deployment efficiency of wireless access point locations.

[0006] In a first aspect, an embodiment of the present application provides a wireless network intelligent planning method, including:

[0007] Obtaining a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned, wherein the three-dimensional grid model is constructed based on three-dimensional point cloud data with material information;

[0008] Using ray tracing technology, wireless signal propagation simulation is performed on the three-dimensional grid model to obtain a signal propagation data set;

[0009] The signal propagation data set is input into a wireless network intelligent planning model, and the wireless network intelligent planning model is used to predict the location of a wireless access point to obtain the location of the wireless access point.

[0010] In a possible implementation, obtaining a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned includes:

[0011] Acquiring three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data of the environment to be planned;

[0012] Performing feature fusion on the three-dimensional point cloud data and the multispectral image data to obtain three-dimensional point cloud data with material information;

[0013] constructing the three-dimensional mesh model based on the three-dimensional point cloud data with material information;

[0014] The electromagnetic parameter distribution is determined based on the multispectral image data and the radar cross-section data.

[0015] In a possible implementation, the performing feature fusion on the three-dimensional point cloud data and the multispectral image data to obtain the three-dimensional point cloud data with material information includes:

[0016] Extracting spectral information from the multispectral image data, and extracting texture information from the RGB image data of the environment to be planned;

[0017] determining material information based on the texture information and the spectral information;

[0018] Determining a point cloud-pixel mapping table based on the three-dimensional point cloud data and the RGB image data;

[0019] Based on the point cloud-pixel mapping table, the material information is mapped to the three-dimensional point cloud data to determine the three-dimensional point cloud data with material information.

[0020] In a possible implementation, constructing the three-dimensional mesh model based on the three-dimensional point cloud data with material information includes:

[0021] Inputting the three-dimensional point cloud data with material information into a three-dimensional mesh construction model, wherein the three-dimensional mesh construction model constructs the three-dimensional mesh model based on the three-dimensional point cloud data with material information, including:

[0022] Inputting the three-dimensional point cloud data with material information into a point cloud processing network, wherein the point cloud processing network obtains three-dimensional point cloud data with normal vectors based on the three-dimensional point cloud data with material information;

[0023] Surface reconstruction is performed based on the three-dimensional point cloud data with normal vectors to obtain the three-dimensional mesh model.

[0024] In a possible implementation, determining the electromagnetic parameter distribution based on the multispectral image data and the radar cross-section data includes:

[0025] Correcting the radar cross-sectional area data based on the three-dimensional point cloud data with material information to obtain corrected radar cross-sectional area data;

[0026] The electromagnetic parameter distribution is inverted based on the corrected radar cross-section data.

[0027] In a possible implementation, before performing feature fusion on the three-dimensional point cloud data and the multispectral image data, the method further includes:

[0028] The three-dimensional point cloud data, the multispectral image data and the radar cross-sectional area data are time synchronized and spatially calibrated.

[0029] In a possible implementation manner, before inputting the signal propagation dataset into the wireless network intelligent planning model, the method further includes:

[0030] Acquire training set data, the training set data including a plurality of training samples, each of the training samples including: a three-dimensional grid model and an electromagnetic parameter distribution, the three-dimensional grid model being constructed based on three-dimensional point cloud data having material information;

[0031] For each training sample, perform the following steps:

[0032] Using the ray tracing technology, a wireless signal propagation simulation is performed on the three-dimensional grid model of the training sample to obtain a signal propagation data set sample;

[0033] Inputting the signal propagation dataset sample into the wireless network intelligent planning model to obtain a wireless access point location prediction result of the training sample;

[0034] Calculating a reward function value based on the wireless access point location prediction result of the training sample, where the reward function is calculated based on signal coverage, signal-to-noise ratio, and co-channel interference indicators and their corresponding weights;

[0035] According to the reward function value, determining whether the reward function value satisfies a preset training stop condition;

[0036] If not, the model parameters of the wireless network intelligent planning model are adjusted, and the wireless network intelligent planning model is continuously trained using the training sample set until the preset training stop condition is met, thereby obtaining a trained wireless network intelligent planning model.

[0037] In a possible implementation, obtaining training set data includes:

[0038] Acquire multiple environmental scene parameters using a parametric modeling tool, and generate multiple basic three-dimensional scenes based on the multiple environmental scene parameters;

[0039] Inputting the plurality of basic three-dimensional scenes into a generative adversarial network, wherein the generative adversarial network generates a plurality of extreme three-dimensional scenes based on the plurality of basic three-dimensional scenes;

[0040] Based on the multiple basic three-dimensional scenes, three-dimensional mesh models corresponding to the multiple basic three-dimensional scenes are generated; and based on the multiple extreme three-dimensional scenes, three-dimensional mesh models corresponding to the multiple extreme three-dimensional scenes are generated.

[0041] In a second aspect, an embodiment of the present application provides a wireless network intelligent planning device, including:

[0042] An acquisition module, configured to acquire a three-dimensional grid model of the environment to be planned and the distribution of electromagnetic parameters, wherein the three-dimensional grid model is constructed based on three-dimensional point cloud data having material information;

[0043] A processing module, configured to simulate wireless signal propagation on a three-dimensional grid model using ray tracing technology to obtain a signal propagation data set;

[0044] The prediction module is configured to input the signal propagation data set into a wireless network intelligent planning model, and predict the location of a wireless access point using the wireless network intelligent planning model to obtain the location of the wireless access point.

[0045] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory storing computer program instructions;

[0046] When the processor executes the computer program instructions, the wireless network intelligent planning method as described in any one of the first aspects is implemented.

[0047] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the wireless network intelligent planning method as described in any one of the first aspects is implemented.

[0048] In a fifth aspect, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the wireless network intelligent planning method as described in any one of the first aspects.

[0049] The embodiments of the present application provide a wireless network intelligent planning method, apparatus, device, storage medium, and product. These methods obtain a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned, then utilize ray tracing technology to simulate wireless signal propagation on the three-dimensional grid model to obtain a signal propagation dataset. Ultimately, the signal propagation dataset is input into the wireless network intelligent planning model, and the wireless network intelligent planning model is used to predict the location of wireless access points. Compared to the prior art, which requires engineers to determine preliminary locations of wireless access points based on their experience, then measure signal strength at various locations in the environment, and continuously adjust the locations of wireless access points based on the signal strength, the present application simulates wireless signal propagation using ray tracing technology to obtain a signal propagation dataset, and then automatically predicts wireless access point locations based on the wireless network intelligent planning model. This eliminates the need for engineers to conduct offline testing and deployment, or to manually search for optimal wireless access point locations through simulation. Instead, the wireless network intelligent planning model automatically predicts wireless access point locations, improving AP location deployment efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0051] Figure 1 A flowchart of a first embodiment of a wireless network intelligent planning method provided by this application;

[0052] Figure 2 A flowchart of a second embodiment of a wireless network intelligent planning method provided by this application;

[0053] Figure 3 Schematic diagram of the connection structure of each sensor;

[0054] Figure 4 A flowchart of a third embodiment of a wireless network intelligent planning method provided by this application;

[0055] Figure 5 A schematic diagram of the structure of a wireless network intelligent planning device provided in this application

[0056] Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in this application. DETAILED DESCRIPTION

[0057] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating the examples of the present application.

[0058] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, the elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or device comprising the elements.

[0059] Conventional technologies require engineers to determine wireless access point locations based on field surveys and their experience. However, complex environments often require engineers to manually adjust the wireless access point locations multiple times to find the most suitable access point, reducing the efficiency of wireless access point deployment. Furthermore, conventional technologies often fail to fully reflect the actual interaction between signals and the actual environment when faced with complex indoor environments with diverse structures and materials, as well as high-frequency signal propagation characteristics. This can lead to insufficient planning accuracy, numerous blind spots, and redundant areas.

[0060] Based on this, the inventive concept of the present application is to provide a wireless network intelligent planning method, apparatus, device, storage medium and product that can automatically determine the location of wireless access points, rather than relying on engineers' experience to determine the location of wireless access points. Therefore, the present application obtains a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned, uses ray tracing technology to simulate wireless signal propagation on the three-dimensional grid model, obtains a signal propagation data set, and uses the wireless network intelligent planning model trained by the present application to predict the location of wireless access points to obtain the final location of wireless access points, thereby significantly improving the efficiency of wireless access point location deployment.

[0061] In order to solve the problems of the prior art, the embodiments of the present application provide a wireless network intelligent planning method, apparatus, device, storage medium and product.

[0062] Figure 1 This is a flow chart of a first embodiment of a wireless network intelligent planning method provided by this application, such as Figure 1 As shown, the method includes the following steps:

[0063] S101: Acquire a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned, where the three-dimensional grid model is constructed based on three-dimensional point cloud data with material information.

[0064] In this embodiment, a 3D mesh model and electromagnetic parameter distribution of the planned environment are obtained. The 3D mesh model is constructed based on 3D point cloud data containing material information. The 3D mesh model is a digital representation of the planned environment, used to simulate the propagation of wireless signals in the environment. The electromagnetic parameter distribution refers to key parameters of the objects in the environment that affect wireless signals, such as dielectric constant and conductivity.

[0065] S102: Using ray tracing technology, perform wireless signal propagation simulation on the three-dimensional grid model to obtain a signal propagation data set.

[0066] In the embodiments of this application, ray tracing technology is used to simulate wireless signal propagation in a three-dimensional grid model to generate a signal propagation dataset. Specifically, ray tracing technology sets ray parameters such as operating frequency band, transmit power, and antenna pattern. Then, based on these ray parameters, it simulates the propagation path and characteristics of wireless signals in a three-dimensional scene model and generates a signal propagation dataset. The signal propagation data includes path loss, multipath delay, and angular spread. Each ray has a corresponding signal propagation data set for each operating frequency band, transmit power, and antenna orientation.

[0067] It's important to explain that ray tracing is a high-precision electromagnetic wave propagation modeling technique. Its core concept is to simulate the propagation of wireless signals in space as rays. Specifically, ray tracing first abstracts wireless signals into a large number of discrete rays based on the principles of electromagnetic wave propagation and geometric optics approximations. It then simulates the interaction of these rays with structures in the environment, such as buildings, obstacles, or terrain, including reflection, refraction, scattering, and diffraction, ultimately accurately calculating parameters such as signal strength, phase, and delay at the receiving point. Compared to traditional empirical propagation models, such as the free-space propagation model, the wall attenuation model, or empirical corrections, ray tracing can more realistically reproduce the characteristics of wireless propagation in complex environments.

[0068] S103: Inputting the signal propagation data set into the wireless network intelligent planning model, predicting the location of the wireless access point through the wireless network intelligent planning model, and obtaining the location of the wireless access point.

[0069] In the embodiments of the present application, a signal propagation dataset is input into a wireless network intelligent planning model, which then predicts the location of wireless access points (APs) to obtain final AP locations. It should be noted that the AP locations obtained by the wireless network intelligent planning model are suitable for the planned environment and can significantly improve signal strength in the planned environment.

[0070] In an embodiment of the present application, a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned are obtained. Ray tracing technology is then used to simulate wireless signal propagation on the three-dimensional grid model to obtain a signal propagation dataset. This signal propagation dataset is then input into a wireless network intelligent planning model, which predicts the locations of wireless access points (APs). Compared to existing techniques that rely on engineers' experience to determine preliminary AP locations, then measure signal strength at various locations in the environment and continuously adjust AP locations based on the signal strength, the present application simulates wireless signal propagation using ray tracing technology to obtain a signal propagation dataset. The AP locations are then automatically predicted based on the AP location using the AP location intelligent planning model. This eliminates the need for engineers to conduct offline testing and deployment, or to manually search for optimal AP locations through simulation. The AP locations are automatically predicted using the AP location intelligent planning model, automating the entire process from data collection to AP location, significantly shortening the network planning cycle and reducing deployment costs.

[0071] Figure 2 This is a flow chart of a second embodiment of a wireless network intelligent planning method provided by this application. Figure 1 Based on the embodiment shown, a specific implementation of step S101 is as follows:

[0072] S201: Acquire three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data of the environment to be planned;

[0073] In an embodiment of the present application, three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data of the planned environment are obtained. In one example, a laser radar is used to scan the planned environment to obtain three-dimensional spatial coordinates to describe the geometric structure of the planned environment, such as the outline and size of walls and furniture, to obtain three-dimensional point cloud data; a multispectral camera is used to collect multi-band reflectivity data of the planned environment, such as visible light and near-infrared bands, to obtain multispectral image data; and a millimeter-wave radar is used to transmit millimeter-wave signals, receive target reflection signals, and calculate radar cross-sectional area data.

[0074] To improve the accuracy of multi-source data fusion and provide accurate data support for the subsequent construction of a 3D grid model and electromagnetic parameter distribution that is more in line with the actual environment, before the 3D point cloud data and multispectral image data are fused, the following steps are also included:

[0075] The three-dimensional point cloud data, multispectral image data and radar cross-section data are time synchronized and spatially calibrated.

[0076] In an embodiment of the present application, time synchronization and spatial calibration are performed on three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data to improve the accuracy of multi-source data fusion and construct a three-dimensional grid model and electromagnetic parameter distribution that is more in line with the actual environment.

[0077] In one example, the specific steps for time synchronization of 3D point cloud data, multispectral image data, and radar cross-section data are as follows:

[0078] Before the LiDAR, multispectral camera and millimeter-wave radar collect data, the local clock of each sensor device is calibrated by software according to the Network Time Protocol (NTP) to ensure that the time when each sensor device collects data is consistent with the standard time provided by the Global Positioning System (GPS). In addition, when each sensor device collects data, the dual synchronization mechanism of Pulse Per Second (PPS) and hardware trigger line is used to ensure that the time when the LiDAR, multispectral camera and millimeter-wave radar collect data is the same. Specifically, the GPS module outputs the PPS pulse signal to the Field Programmable Gate Array (FPGA) controller, and the FPGA controller controls the sampling of each sensor device at the same time through the hardware trigger line, thereby achieving synchronization of the sampling instant at the hardware level, further improving the accuracy of time alignment, such as Figure 3 As shown, Figure 3 Schematic diagram of the connection structure of each sensor.

[0079] Then, after each sensor collects data, the collected data is interpolated according to the timestamp interpolation algorithm to compensate for the slight delay of the hardware trigger. After achieving high-precision time synchronization sampling through GPS / PPS pulse signals and hardware trigger lines, since the sampling time of each sensor recording data is the local time of each sampling moment (i.e., the timestamp of the moment the hardware trigger line is triggered), due to factors such as the internal response speed of the sensor and communication delay, there is still a slight time deviation, generally at the sub-millimeter level. Therefore, in order to further improve the accuracy of time alignment, after each sensor collects data, according to the timestamp interpolation algorithm, by recording the precise timestamps before and after each sensor sampling, combined with the interpolation formula, the time of each sampling point is accurately mapped to a unified GPS time reference, thereby compensating for the slight delay in the hardware trigger process and achieving higher-precision time alignment.

[0080] In one example, the specific steps for spatial calibration of 3D point cloud data, multispectral image data, and radar cross-section data are as follows:

[0081] Unify 3D point cloud data, multispectral image data, and radar cross-section data into the same coordinate system to facilitate subsequent data fusion and processing. For example, multispectral image data and radar cross-section data can be converted into the coordinate system of 3D point cloud data. For example, the coordinate transformation relationship between radar cross-section data and 3D point cloud data can be determined based on corner reflector calibration, and then the radar cross-section data can be converted into the 3D point cloud data coordinate system based on the coordinate transformation relationship. The multispectral image data can also be converted into the 3D point cloud data coordinate system based on the extended Zhang Zhengyou calibration method using a multispectral checkerboard.

[0082] In an embodiment of the present application, three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data are time-synchronized and spatially calibrated to achieve high-precision alignment of multi-source data, solving the problem of low multi-source data fusion accuracy due to large multi-source data alignment errors, and providing accurate data support for the subsequent construction of a three-dimensional grid model and electromagnetic parameter distribution that is more in line with the actual environment.

[0083] S202: Fusing the three-dimensional point cloud data and the multispectral image data to obtain three-dimensional point cloud data with material information.

[0084] In an embodiment of the present application, feature fusion is performed on the three-dimensional point cloud data and the multispectral image data, and the material and color information in the multispectral image data is fused into the three-dimensional point cloud data, so that each point cloud data has not only geometric position information but also material information, providing a data basis for the subsequent construction of a more accurate and realistic three-dimensional mesh model.

[0085] In order to construct a three-dimensional grid model that better fits the environment, a specific implementation of step S202 may further include the following steps:

[0086] Spectral information is extracted from multispectral image data, and texture information is extracted from RGB image data of the environment to be planned.

[0087] In this embodiment, spectral information is extracted from multispectral image data. A multispectral camera collects reflectance data in multiple bands, with each pixel corresponding to a spectral vector. Texture information is extracted from RGB image data of the planned environment using a texture analysis algorithm or a deep learning model.

[0088] It should be noted that the RGB image data of the planned environment is collected by a visible light camera, and before extracting spectral information from the multispectral image data and texture information from the RGB image data of the planned environment, the RGB image data is first time-synchronized and spatially calibrated with the 3D point cloud data, multispectral image data, and radar cross-sectional area data. The time synchronization process is the same as the specific steps for time synchronization of the 3D point cloud data, multispectral image data, and radar cross-sectional area data described above, and will not be described in detail here. During spatial calibration, the extrinsic parameter matrices of the RGB camera and lidar can be accurately calculated based on the Zhang Zhengyou calibration method using a checkerboard calibration plate, and the RGB image data can be converted into the 3D point cloud data coordinate system based on the extrinsic parameter matrix.

[0089] Determine material information based on texture information and spectral information.

[0090] In an embodiment of the present application, material information is determined based on texture information extracted from RGB image data and spectral information extracted from multispectral image data. In one example, material information is determined based on a multimodal fusion model that combines texture information and spectral information. For example, the spectral information and texture information are input into a hybrid model of ResNet-50 and Transformer, which outputs a material category probability. For example, if the material category determined by certain texture information and spectral information is: metal 70%, glass 30%, then the material category with the highest probability is selected as the material information, that is, metal is determined as the material information.

[0091] Based on the 3D point cloud data and the RGB image data, a point cloud-pixel mapping table is determined.

[0092] In an embodiment of the present application, a point cloud-pixel mapping table is determined based on three-dimensional point cloud data and RGB image data. In one example, the three-dimensional point cloud data and RGB image data are matched in real time based on an iterative closest point algorithm to generate a point cloud-pixel mapping table to ensure accurate correspondence between geometry and texture information.

[0093] Based on the point cloud-pixel mapping table, the material information is mapped to the three-dimensional point cloud data to determine the three-dimensional point cloud data with material information.

[0094] In an embodiment of the present application, a point cloud-pixel mapping table is used to find the corresponding pixel in the RGB image for each three-dimensional point cloud, and then the material information of the pixel is obtained, and the corresponding material information is mapped to the point cloud data, and finally three-dimensional point cloud data with material information is obtained.

[0095] In one example, determining three-dimensional point cloud data with material information may also include: performing feature fusion on the three-dimensional point cloud data and multispectral image data based on a multimodal fusion network driven by an attention mechanism to obtain three-dimensional point cloud data with material information.

[0096] In an embodiment of the present application, spectral information is first extracted from multispectral image data, and texture information is extracted from RGB image data of the environment to be planned. Based on the texture information and spectral information, material information is determined. Based on the three-dimensional point cloud data and the RGB image data, a point cloud-pixel mapping table is determined. Finally, based on the point cloud-pixel mapping table, the material information is mapped to the three-dimensional point cloud data to determine the three-dimensional point cloud data with material information. The three-dimensional point cloud data with material information can be used to construct a more accurate and more practical three-dimensional grid model, so that when simulating wireless signal propagation according to ray tracing technology, it is more in line with the wireless signal propagation in the actual environment, so that the constructed three-dimensional grid model has richer environmental information, so that the final prediction of the wireless access point location according to the wireless network intelligent planning model is more accurate.

[0097] S203: Constructing a three-dimensional mesh model based on the three-dimensional point cloud data with material information.

[0098] In an embodiment of the present application, a three-dimensional mesh model is constructed based on three-dimensional point cloud data with material information.

[0099] In one example, a specific implementation method for constructing a 3D mesh model based on 3D point cloud data with material information is as follows:

[0100] Input the 3D point cloud data with material information into the 3D mesh construction model. The 3D mesh construction model is constructed based on the 3D point cloud data with material information, including:

[0101] The three-dimensional point cloud data with material information is input into the point cloud processing network, and the point cloud processing network obtains the three-dimensional point cloud data with normal vectors based on the three-dimensional point cloud data with material information.

[0102] In an embodiment of the present application, three-dimensional point cloud data with material information is input into a point cloud processing network, for example, an improved Point Transformer network. The point cloud processing network captures the global context of the three-dimensional point cloud data with material information through a self-attention mechanism and generates three-dimensional point cloud data with normal vectors.

[0103] In one example, before inputting 3D point cloud data with material information into a point cloud processing network, the data is preprocessed. Outlier noise points are removed using a statistical filtering algorithm, and the density of the material-rich 3D point cloud data is reduced using a voxel grid downsampling algorithm. This reduces redundant data within the material-rich 3D point cloud data, improving the computational efficiency of the point cloud processing network. The point cloud processing network then generates dense 3D point cloud data with normal vectors to improve the accuracy of the subsequent 3D mesh model. By first removing explicit noise and redundancy using lightweight operations, and then recovering implicit features and details through deep learning, the resulting reconstruction achieves higher quality than directly processing the raw data.

[0104] Surface reconstruction is performed based on 3D point cloud data with normal vectors to obtain a 3D mesh model.

[0105] In an embodiment of the present application, surface reconstruction is performed based on 3D point cloud data with normal vectors. For example, the 3D point cloud data with normal vectors is input into a Poisson surface reconstruction algorithm. The Poisson surface reconstruction algorithm converts the 3D point cloud data with normal vectors into a closed 3D mesh model. The 3D mesh model supports industry standard formats such as OBJ and GLFT. It will be understood that because the 3D mesh model is constructed based on 3D point cloud data with material information, the constructed 3D mesh model contains material information.

[0106] S204: Determine electromagnetic parameter distribution based on the multispectral image data and radar cross-section data.

[0107] In an embodiment of the present application, electromagnetic parameter distribution is determined based on multispectral image data and radar cross-section data, and the electromagnetic parameter distribution includes dielectric constant distribution and conductivity distribution.

[0108] In one example, a specific implementation method for determining electromagnetic parameter distribution based on multispectral image data and radar cross-section data is as follows:

[0109] The radar cross-sectional area data is corrected based on the three-dimensional point cloud data with material information to obtain the corrected radar cross-sectional area data.

[0110] In an embodiment of the present application, radar cross-section data is corrected based on 3D point cloud data with material information. For example, the material-labeled 3D point cloud data and radar cross-section data are fused using a Bayesian network to dynamically correct the radar cross-section data and generate corrected radar cross-section data. Correcting the radar cross-section data with the 3D point cloud data with material information can eliminate errors in the radar cross-section data caused by environmental noise or multipath effects, making the corrected radar cross-section data more consistent with actual material characteristics.

[0111] Based on the corrected radar cross-section data, the electromagnetic parameter distribution is inverted.

[0112] In an embodiment of the present application, the electromagnetic parameter distribution is inverted based on the corrected radar cross-section data. For example, initial values ​​are first set for the dielectric constant and conductivity based on the COST 2100 material database. COST 2100 is a widely used material database that contains electromagnetic parameters of many common materials. It provides prior knowledge for the inversion process, helps to narrow the search space and improve the inversion efficiency. Then, the Fresnel equation and the initial dielectric constant and conductivity are used to calculate the reflection characteristics of the target object to the radar wave to obtain simulated radar cross-section data. The simulated radar cross-section data is compared with the corrected radar cross-section data, and the error between the two is calculated. The values ​​of the dielectric constant and conductivity are adjusted using an iterative optimization algorithm according to the error result until the error reaches a preset threshold to obtain the final values ​​of the dielectric constant and conductivity. The electromagnetic parameter values ​​are adjusted point by point to obtain the final electromagnetic parameter distribution.

[0113] In an embodiment of the present application, the radar cross-sectional area data is corrected based on the three-dimensional point cloud data with material information to obtain the corrected radar cross-sectional area data, and then the electromagnetic parameter distribution is inverted based on the corrected radar cross-sectional area data. The radar cross-sectional area data is corrected according to the material information to make it closer to the radar cross-sectional area data in the actual environment, and then the electromagnetic parameter distribution is inverted based on the corrected radar cross-sectional area data to make the obtained electromagnetic parameter distribution closer to the electromagnetic parameter distribution of the actual environment.

[0114] In an embodiment of the present application, three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data of the environment to be planned are acquired, and then feature fusion is performed on the three-dimensional point cloud data and the multispectral image data to obtain three-dimensional point cloud data with material information. A three-dimensional grid model is constructed based on the three-dimensional point cloud data with material information, and electromagnetic parameter distribution is determined based on the multispectral image data and radar cross-sectional area data. By deeply fusing multi-sensor data, the three-dimensional information of the environment and the material characteristics of the scatterers are accurately captured, fully reflecting the true interaction between the signal and the actual environment. This provides a precise three-dimensional grid model and electromagnetic parameter environment for subsequent prediction of wireless access point locations based on a wireless network intelligent planning model, further improving the accuracy and efficiency of wireless access point location prediction.

[0115] Figure 4 This is a flow chart of a third embodiment of a wireless network intelligent planning method provided by this application. Figure 1 Based on the embodiment shown, before step S103, the method further includes:

[0116] S401: Acquire training set data, where the training set data includes multiple training samples. Each training sample includes a three-dimensional grid model and an electromagnetic parameter distribution. The three-dimensional grid model is constructed based on three-dimensional point cloud data with material information.

[0117] To improve the generalization capability of the wireless network intelligent planning model, a simulation dataset covering diverse scenarios was constructed, including:

[0118] A plurality of environmental scene parameters are obtained according to a parametric modeling tool, and a plurality of basic three-dimensional scenes are generated based on the plurality of environmental scene parameters.

[0119] In an embodiment of the present application, multiple environmental scene parameters are acquired using a parametric modeling tool, such as a Blender script, and multiple basic 3D scenes are generated based on the multiple environmental scene parameters. In one example, a random range of wall thickness is defined in the Blender script, for example, 10-30 cm. For each generated basic 3D scene, a wall thickness value is randomly selected. To ensure that the furniture is reasonably distributed within the basic 3D scene, neither too dense nor too sparse, the furniture distribution within each 3D scene is determined using a uniform distribution sampling algorithm. Material types are randomly extracted from the NLCD material library and assigned to objects such as walls and furniture.

[0120] Multiple basic three-dimensional scenes are input into the adversarial generative network, which generates multiple extreme three-dimensional scenes based on the multiple basic three-dimensional scenes.

[0121] In an embodiment of the present application, multiple basic three-dimensional scenes are input into a generative adversarial network, and the generative adversarial network generates multiple extreme three-dimensional scenes based on the multiple basic three-dimensional scenes, such as extreme three-dimensional scenes with high-density obstacles or heterogeneous structures.

[0122] Based on multiple basic three-dimensional scenes, three-dimensional grid models corresponding to the multiple basic three-dimensional scenes are generated; and based on multiple extreme three-dimensional scenes, three-dimensional grid models corresponding to the multiple extreme three-dimensional scenes are generated.

[0123] In an embodiment of the present application, based on multiple basic three-dimensional scenes, they are discretized into grids with a resolution of 0.1 meters to generate three-dimensional grid models corresponding to the multiple basic three-dimensional scenes, and based on multiple extreme three-dimensional scenes, they are discretized into grids with a resolution of 0.1 meters to generate three-dimensional grid models corresponding to the multiple extreme three-dimensional scenes.

[0124] For each training sample, perform the following steps:

[0125] S402: Using ray tracing technology, perform wireless signal propagation simulation on the three-dimensional grid model of the training sample to obtain a signal propagation data set sample.

[0126] In an embodiment of the present application, a 3D mesh model is imported into a ray tracing engine, and ray tracing technology is used to simulate wireless signal propagation on the 3D mesh model of the training sample, thereby obtaining a sample signal propagation dataset. In one example, the 3D mesh model is imported into the ray tracing engine, and the electromagnetic parameters of obstacles such as walls and furniture are defined based on the electromagnetic parameter distribution corresponding to the 3D mesh model. Then, by setting the operating frequency band of the wireless access point location (e.g., 5 GHz, 28 GHz), the transmit power, and the antenna pattern, wireless signal propagation is simulated to obtain a sample signal propagation dataset. Each signal propagation data includes path loss, multipath delay, and angular spread.

[0127] S403: Input the signal propagation dataset sample into the wireless network intelligent planning model to obtain the wireless access point location prediction result of the training sample.

[0128] In an embodiment of the present application, a signal propagation dataset sample is input into a wireless signal network intelligent planning model to obtain a predicted wireless access point location for the training sample. In one example, the wireless network intelligent planning model is an agent that uses ray tracing technology to simulate wireless signal propagation. After obtaining the signal propagation dataset sample, the agent calculates the signal strength and interference level of each grid point based on the signal propagation dataset sample. The agent then preliminarily explores the location of the wireless access point based on the signal strength and interference level of each grid point to obtain a predicted wireless access point location for the training sample.

[0129] S404: Calculate a reward function value based on the wireless access point location prediction result of the training sample. The reward function is calculated based on signal coverage, signal-to-noise ratio, and co-channel interference indicators and their corresponding weights.

[0130] In this embodiment of the present application, the agent calculates a reward function value based on the wireless access point location prediction results of the training sample. The reward function value is calculated based on the signal coverage rate, signal-to-noise ratio, and co-channel interference index and their corresponding weights. The reward function formula is as follows:

[0131] R=αCoverage+βSNR-γInterference (1)

[0132] Among them, R is the reward function value, Coverage is the signal coverage, SNR is the signal-to-noise ratio, Interference is the co-channel interference index, α is the weight corresponding to the signal coverage, β is the weight corresponding to the signal-to-noise ratio, and γ is the weight corresponding to the co-channel interference index.

[0133] S405: According to the reward function value, determine whether the reward function value meets the preset training stop condition.

[0134] In an embodiment of the present application, based on the reward function value, it is determined whether the reward function value meets a preset training stop condition, wherein the preset training stop condition is: the reward function value curve converges stably.

[0135] S406: If not, adjust the model parameters of the wireless network intelligent planning model, and continue to train the wireless network intelligent planning model using the training sample set until a preset training stop condition is met, thereby obtaining a trained wireless network intelligent planning model.

[0136] In an embodiment of the present application, if the preset training stop condition is not met, the model parameters of the wireless network intelligent planning model are adjusted based on the reward function value. In one example, the model parameters of the wireless network intelligent planning model are adjusted using a proximal policy optimization algorithm. The proximal policy optimization algorithm adjusts the model parameters of the wireless network intelligent planning model based on the reward function value. The proximal policy optimization algorithm avoids drastic policy changes by limiting the amplitude of each policy update, thereby enhancing the stability of the wireless network intelligent planning model algorithm. The wireless network intelligent planning model after parameter adjustment then continues to predict the location of the wireless access point. The newly predicted wireless access point location is then used to determine whether the preset training stop condition is met. This is done until the preset training stop condition is met, thereby obtaining a trained wireless network intelligent planning model.

[0137] In an embodiment of the present application, a wireless network intelligent planning model is trained to obtain the optimal wireless access point locations for multiple training set samples. A reinforcement learning agent is then used to automatically search for the optimal wireless access point locations in a simulation environment and record their signal strength, coverage, and signal-to-noise ratio as gold standard labels. The resulting scene geometry model, material distribution, electromagnetic parameters, and corresponding signal propagation dataset samples and optimal wireless access point location prediction results are then stored in a hierarchical HDF5 format. This creates a large-scale scene-layout pair covering 20 building types, including offices, factories, and residences. This provides sufficient samples for subsequent model training, thereby improving the adaptability and robustness of the wireless network intelligent planning model.

[0138] Figure 5 This is a schematic diagram of the structure of a wireless network intelligent planning device provided by this application, such as Figure 5 As shown, the wireless network intelligent planning device 500 provided in the embodiment of the present application includes:

[0139] The acquisition module 501 is used to acquire a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned. The three-dimensional grid model is constructed based on three-dimensional point cloud data with material information.

[0140] The processing module 502 is configured to use ray tracing technology to perform wireless signal propagation simulation on the three-dimensional grid model to obtain a signal propagation data set.

[0141] The prediction module 503 is configured to input the signal propagation data set into the wireless network intelligent planning model, and predict the location of the wireless access point using the wireless network intelligent planning model to obtain the location of the wireless access point.

[0142] In a possible implementation, the apparatus may further include:

[0143] An acquisition module, configured to acquire three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data of the environment to be planned;

[0144] The fusion module is used to fuse the 3D point cloud data and multispectral image data to obtain 3D point cloud data with material information;

[0145] The construction module constructs a 3D mesh model based on 3D point cloud data with material information;

[0146] The determination module determines the electromagnetic parameter distribution based on the multispectral image data and the radar cross-section data.

[0147] In a possible implementation, the apparatus may further include:

[0148] An extraction module, configured to extract spectral information from the multispectral image data and texture information from the RGB image data of the environment to be planned;

[0149] a determination module, which determines material information based on the texture information and the spectral information;

[0150] A determination module, which determines a point cloud-pixel mapping table based on the three-dimensional point cloud data and the RGB image data;

[0151] A determination module maps the material information to the three-dimensional point cloud data based on the point cloud-pixel mapping table to determine the three-dimensional point cloud data with material information.

[0152] In a possible implementation, the apparatus may further include:

[0153] An input module inputs the three-dimensional point cloud data with material information into a point cloud processing network, wherein the point cloud processing network obtains three-dimensional point cloud data with normal vectors based on the three-dimensional point cloud data with material information;

[0154] A processing module is used to perform surface reconstruction based on the three-dimensional point cloud data with normal vectors to obtain the three-dimensional mesh model.

[0155] In a possible implementation, the apparatus may further include:

[0156] a correction module, which corrects the radar cross-sectional area data based on the three-dimensional point cloud data with material information to obtain corrected radar cross-sectional area data;

[0157] The inversion module inverts the electromagnetic parameter distribution based on the corrected radar cross-section data.

[0158] In a possible implementation, the apparatus may further include:

[0159] A processing module is used to perform time synchronization and spatial calibration on the three-dimensional point cloud data, the multispectral image data and the radar cross-sectional area data.

[0160] In a possible implementation, the apparatus may further include:

[0161] An acquisition module is configured to acquire training set data, wherein the training set data includes a plurality of training samples, each of the training samples including: a three-dimensional grid model and an electromagnetic parameter distribution, wherein the three-dimensional grid model is constructed based on three-dimensional point cloud data having material information;

[0162] For each training sample, perform the following steps:

[0163] a processing module, configured to perform wireless signal propagation simulation on the three-dimensional grid model of the training sample using the ray tracing technology to obtain a signal propagation data set sample;

[0164] a processing module, configured to input the signal propagation dataset sample into the wireless network intelligent planning model to obtain a wireless access point location prediction result of the training sample;

[0165] A processing module calculates a reward function value based on the wireless access point location prediction result of the training sample, where the reward function is calculated based on signal coverage, signal-to-noise ratio, and co-channel interference indicators and their corresponding weights;

[0166] A processing module is used to determine whether the reward function value satisfies a preset training stop condition based on the reward function value; if not, adjust the model parameters of the wireless network intelligent planning model, and continue training the wireless network intelligent planning model using the training sample set until the preset training stop condition is met, thereby obtaining a trained wireless network intelligent planning model.

[0167] In a possible implementation, the apparatus may further include:

[0168] A processing module, configured to obtain a plurality of environmental scene parameters according to a parametric modeling tool, and generate a plurality of basic three-dimensional scenes based on the plurality of environmental scene parameters;

[0169] a processing module, configured to input the plurality of basic three-dimensional scenes into a generative adversarial network, wherein the generative adversarial network generates a plurality of extreme three-dimensional scenes based on the plurality of basic three-dimensional scenes;

[0170] The processing module generates three-dimensional mesh models corresponding to the multiple basic three-dimensional scenes based on the multiple basic three-dimensional scenes, and generates three-dimensional mesh models corresponding to the multiple extreme three-dimensional scenes based on the multiple extreme three-dimensional scenes.

[0171] Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in this application.

[0172] The electronic device may include a processor 601 and a memory 602 storing computer program instructions.

[0173] Specifically, the processor 601 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0174] The memory 602 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 602 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 602 may include removable or non-removable (or fixed) media. Where appropriate, the memory 602 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 602 is a non-volatile solid-state memory.

[0175] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical or other physical / tangible memory storage devices. Thus, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0176] The processor 601 implements any one of the wireless network intelligent planning methods in the above embodiments by reading and executing computer program instructions stored in the memory 302 .

[0177] In one example, the electronic device may further include a communication interface 603 and a bus 610. Figure 6 As shown, the processor 601, the memory 602, and the communication interface 603 are connected via a bus 610 and communicate with each other.

[0178] The communication interface 603 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0179] Bus 610 includes hardware, software or both, and the parts of online data flow metering equipment are coupled to each other. For example, but not limitation, bus can include accelerated graphics port (AGP) or other graphics bus, enhanced industry standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industry standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 610 can include one or more buses. Although the present application embodiment describes and shows specific bus, the application considers any suitable bus or interconnection.

[0180] In addition, in conjunction with the wireless network intelligent planning method in the above embodiments, the present application embodiment may provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any of the wireless network intelligent planning methods in the above embodiments is implemented.

[0181] An embodiment of the present application further provides a computer program product, including a computer program, which, when executed, implements any one of the methods for intelligent planning of wireless networks in the above embodiments.

[0182] It should be understood that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present application.

[0183] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium that can store or transmit information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memories, erasable ROMs (EROMs), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0184] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps. In other words, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0185] Aspects of the present application have been described above with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed via the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. This processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It is also understood that each box in the block diagram and / or the flowchart and the combination of the boxes in the block diagram and / or the flowchart can also be implemented by the dedicated hardware that performs the specified function or action, or can be implemented by the combination of dedicated hardware and computer instructions.

[0186] The above description is only a specific embodiment of the present application. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the scope of protection of the present application.

Claims

1. A wireless network intelligent planning method, characterized in that: include: Obtaining a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned, wherein the three-dimensional grid model is constructed based on three-dimensional point cloud data with material information; Using ray tracing technology, wireless signal propagation simulation is performed on the three-dimensional grid model to obtain a signal propagation data set; The signal propagation data set is input into a wireless network intelligent planning model, and the wireless network intelligent planning model is used to predict the location of a wireless access point to obtain the location of the wireless access point.

2. The method according to claim 1, characterized in that The step of obtaining a three-dimensional grid model and electromagnetic parameter distribution of the environment to be planned includes: Acquiring three-dimensional point cloud data, multispectral image data, and radar cross-sectional area data of the environment to be planned; Performing feature fusion on the three-dimensional point cloud data and the multispectral image data to obtain three-dimensional point cloud data with material information; constructing the three-dimensional mesh model based on the three-dimensional point cloud data with material information; The electromagnetic parameter distribution is determined based on the multispectral image data and the radar cross-section data.

3. The method according to claim 2, characterized in that The step of performing feature fusion on the three-dimensional point cloud data and the multispectral image data to obtain three-dimensional point cloud data with material information includes: Extracting spectral information from the multispectral image data, and extracting texture information from the RGB image data of the environment to be planned; determining material information based on the texture information and the spectral information; Determining a point cloud-pixel mapping table based on the three-dimensional point cloud data and the RGB image data; Based on the point cloud-pixel mapping table, the material information is mapped to the three-dimensional point cloud data to determine the three-dimensional point cloud data with material information.

4. The method according to claim 2, characterized in that The constructing of the three-dimensional mesh model based on the three-dimensional point cloud data with material information includes: Inputting the three-dimensional point cloud data with material information into a three-dimensional mesh construction model, wherein the three-dimensional mesh construction model constructs the three-dimensional mesh model based on the three-dimensional point cloud data with material information, including: Inputting the three-dimensional point cloud data with material information into a point cloud processing network, wherein the point cloud processing network obtains three-dimensional point cloud data with normal vectors based on the three-dimensional point cloud data with material information; Surface reconstruction is performed based on the three-dimensional point cloud data with normal vectors to obtain the three-dimensional mesh model.

5. The method according to claim 2, characterized in that The determining of the electromagnetic parameter distribution based on the multispectral image data and the radar cross-section data includes: Correcting the radar cross-sectional area data based on the three-dimensional point cloud data with material information to obtain corrected radar cross-sectional area data; The electromagnetic parameter distribution is inverted based on the corrected radar cross-section data.

6. The method according to claim 2, characterized in that Before performing feature fusion on the three-dimensional point cloud data and the multispectral image data, the method further includes: The three-dimensional point cloud data, the multispectral image data and the radar cross-sectional area data are time synchronized and spatially calibrated.

7. The method according to claim 1, characterized in that Before inputting the signal propagation data set into the wireless network intelligent planning model, the method further includes: Acquire training set data, the training set data including a plurality of training samples, each of the training samples including: a three-dimensional grid model and an electromagnetic parameter distribution, the three-dimensional grid model being constructed based on three-dimensional point cloud data having material information; For each training sample, perform the following steps: Using the ray tracing technology, a wireless signal propagation simulation is performed on the three-dimensional grid model of the training sample to obtain a signal propagation data set sample; Inputting the signal propagation dataset sample into the wireless network intelligent planning model to obtain a wireless access point location prediction result of the training sample; Calculating a reward function value based on the wireless access point location prediction result of the training sample, where the reward function is calculated based on signal coverage, signal-to-noise ratio, and co-channel interference indicators and their corresponding weights; According to the reward function value, determining whether the reward function value satisfies a preset training stop condition; If not, the model parameters of the wireless network intelligent planning model are adjusted, and the wireless network intelligent planning model is continuously trained using the training set until the preset training stop condition is met, thereby obtaining a trained wireless network intelligent planning model.

8. The method according to claim 7, characterized in that The obtaining of training set data includes: Acquire multiple environmental scene parameters using a parametric modeling tool, and generate multiple basic three-dimensional scenes based on the multiple environmental scene parameters; Inputting the plurality of basic three-dimensional scenes into a generative adversarial network, wherein the generative adversarial network generates a plurality of extreme three-dimensional scenes based on the plurality of basic three-dimensional scenes; Based on the multiple basic three-dimensional scenes, three-dimensional mesh models corresponding to the multiple basic three-dimensional scenes are generated; and based on the multiple extreme three-dimensional scenes, three-dimensional mesh models corresponding to the multiple extreme three-dimensional scenes are generated.

9. A wireless network intelligent planning device, characterized in that: include: An acquisition module, configured to acquire a three-dimensional grid model of the environment to be planned and the distribution of electromagnetic parameters, wherein the three-dimensional grid model is constructed based on three-dimensional point cloud data having material information; A processing module is used to simulate wireless signal propagation on a three-dimensional grid model using ray tracing technology to obtain a signal propagation data set; The prediction module is configured to input the signal propagation data set into a wireless network intelligent planning model, and predict the location of a wireless access point using the wireless network intelligent planning model to obtain the location of the wireless access point.

10. An electronic device, characterized in that: The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the wireless network intelligent planning method according to any one of claims 1 to 8 is implemented.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the wireless network intelligent planning method according to any one of claims 1 to 8 is implemented.

12. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the wireless network intelligent planning method according to any one of claims 1 to 8.

Images